DCT-SVD Domain Watermarking Algorithm

The DCT-SVD domain watermarking algorithm is a digital watermarking technique that integrates Discrete Cosine Transform (DCT) and Singular Value Decomposition (SVD), primarily used to embed watermark information into the frequency domain of carrier images to enhance watermark concealment and attack resistance.

The core principle involves transforming images from spatial to frequency domain using DCT, selecting mid-frequency or high-frequency coefficients for watermark embedding to minimize visible impact on image quality. In code implementation, this typically involves applying block-based DCT transformation (using functions like dct2() in MATLAB) and strategically selecting coefficient ranges. Subsequently, SVD decomposition is performed on selected frequency domain blocks (via functions like svd()), where watermark information is embedded by modifying singular values in the diagonal matrix. Due to SVD's mathematical stability, the watermark remains highly extractable even after compression, cropping, or noise attacks.

The algorithm workflow generally includes watermark preprocessing, DCT transformation, block-based SVD decomposition, watermark embedding, and inverse transformation reconstruction. Key implementation steps involve partitioning host images into blocks, applying dual-domain transformations, and controlling embedding strength through scaling factors. This method shows significant potential in copyright protection and authentication applications, effectively balancing concealment and robustness requirements.